The conditions in long-distance offshore areas are complex, and surface wind speeds are highly susceptible to the influence of mesoscale oceanic events. The resulting anomalous data points and bump events will decrease the accuracy of ramp-up detection, affecting the short-term forecasting precision of offshore wind power in long-distance sea areas. Therefore, a short-term forecasting method for offshore wind power in long-distance sea areas is proposed, which simultaneously considers ramp-up events and long-distance sea meteorological factors. Firstly, an improved parameter and resolution adaptive algorithm （PRAA） based on state marker and sliding window is designed to detect ramp-up events and extract features. Secondly, the correlation of multiple factors such as wind speed, wind direction and temperature in the long-distance offshore is analyzed to expand the dimension of the feature samples of the meteorological factors, and the potential features are deeply explored by principal component analysis (PCA). Finally, based on the measured data of a domestic offshore wind farm, the light gradient boosting machine (LightGBM) considering ramp-up and meteorological factors in long-distance sea areas is used to complete the short-term prediction of long-distance offshore wind power. Simulation results verify the effectiveness of the proposed method.